World Families Forums - The "randomness" of the STR Mutational process

Welcome, Guest. Please login or register.
December 25, 2014, 10:00:57 AM
Home Help Search Login Register

+  World Families Forums
|-+  General Forums - Note: You must Be Logged In to post. Anyone can browse.
| |-+  R1b General (Moderator: rms2)
| | |-+  The "randomness" of the STR Mutational process
« previous next »
Pages: [1] 2 3 Go Down Print
Author Topic: The "randomness" of the STR Mutational process  (Read 2474 times)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« on: June 09, 2012, 09:51:42 AM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process. 

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #1 on: June 09, 2012, 10:09:08 AM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process. 

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?

Are you replying on Rootsweb?
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #2 on: June 09, 2012, 10:26:02 AM »

No.  I mailed the person asking the questions the data set.
Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #3 on: June 09, 2012, 10:00:47 PM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process.  

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?

Are you replying on Rootsweb?
No.  I mailed the person asking the questions the data set.

I recommend you put your questions out directly to the people you disagree with.  To simply post "in absentia" comments here is of no value.  Why should we think your points carry any value if you won't argue your position directly and publicly?

I don't really care, other than I don't want a newbie to read your comments without understanding the circumstances.
« Last Edit: June 09, 2012, 10:03:21 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
Arch Y.
Old Hand
****
Offline Offline

Posts: 292


« Reply #4 on: June 10, 2012, 02:57:57 AM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process. 

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?

The link takes me to a page that doesn't exist.

Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #5 on: June 10, 2012, 08:05:25 AM »

It takes you to a list.  select genetic genealogy in the right column.  what you want is the alphabetized list.  Select g.  go down list to genetic genealogy.  it'll give you a table and the first entry is Y STR freq.
Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #6 on: June 10, 2012, 08:08:13 AM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process.  

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?

Are you replying on Rootsweb?
No.  I mailed the person asking the questions the data set.

I recommend you put your questions out directly to the people you disagree with.  To simply post "in absentia" comments here is of no value.  Why should we think your points carry any value if you won't argue your position directly and publicly?

I don't really care, other than I don't want a newbie to read your comments without understanding the circumstances.
  This is a lot of bull mike.  You have your agenda, which I don't happen to agree with and would also be concerned if a newbie read it.  This will sort itself out in time.  Going against the prevailing wind is slow and tedious and admittedly frustrating.  This is your board.  Do with it as you like.
Logged
Jean M
Guru
*****
Offline Offline

Posts: 1253


« Reply #7 on: June 10, 2012, 12:09:06 PM »

I'm getting a little tired of ad hominem attacks here. Mike was simply pointing out that a post on one forum replying to a discussion on another is out of place. It achieves nothing if your desire is to be part of the debate. It cannot be evaluated by readers here without the context. I personally have no interest in STRs at all, so I have no horse in this field.

« Last Edit: June 10, 2012, 12:11:03 PM by Jean M » Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #8 on: June 10, 2012, 12:33:13 PM »

I meant no Ad Hominem attack.  I also agree that this Forum can stand on its own and present its own opinions.  That said, I have heard time and again that this person or that person on rootsweb said this or presented this set of data and ergo, that makes it correct. All of Mikes work is based on one of the rootsweb gurus; Mike has greatly added to our knowledge though by assembling tables of entries such as R-L21 and I applaud him for his efforts, but I'm not convinced we have the right model yet.

I do have an interest in STR's and their meaning.  But, I also think it is premature to predict movements of people at this time when we don't understand the mutational process.  Variance analysis, as it is practiced now, may or may not be correct.  The out of Africa scenario is being questioned.  What do we really know from analyzing STR patterns that is fully accepted and proven by other applicable technologies, such as linguistics, archaeology, history whatever?

I believe we should be in the collecting data mode, not the theoretical mode of predicting the impact of something we don't fully understand, especially over long periods of time. 

Finally, I would recommend everyone read N. Talebs book, the Black Swan.  It descredits the use of the Gaussian in the Financial industry and points out many errors in logic based on the premise that "everything can be described by a Gaussian curve"  JMHO>
Logged
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #9 on: June 10, 2012, 01:00:04 PM »

I must confess I am losing my confidence in age estimates based on STRs. I think they have some value, but it's limited. The problem is that there is nothing in y-dna itself that really tells you how old it is.

Correct me if I am wrong, but wasn't Zhivotovsky's method designed specifically for a study of the Maoris in New Zealand? Am I remembering that correctly? Didn't he feel it was necessary to calibrate the mutation rates according to what was known archaeologically or historically about the arrival of the Maoris in New Zealand? Given that, is it safe to apply Zhivotovsky to situations where no such calibrating data exist?

Germline rates seem safer overall to me, since they are based on empirical studies of father-to-son mutations; but even they have come up short, for example when, based on TMRCA calculations, a number of people claimed E-V13 didn't expand into Europe until the Bronze Age; that has since gone belly-up with the discovery of Neolithic E-V13 in Catalonia.

Zhivotovsky probably produces ages that are way too old. Germline rates may lead to underestimates. Keep doing both and arguing about what they mean. That is fine.

But it seems to me the real hope for some degree of certainty is the expansion of ancient y-dna finds.



Logged

ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #10 on: June 10, 2012, 02:04:44 PM »

I believed he used some Romani gypsies and one other data set also.  Think of the problem like this.  You have a data set, you have an observable number of mutations, then how do you predict mutation rates from this set of data.  I believe thats what he tried to do.  He couldn't count hidden mutations; I don't think he was aware of multisteps etc. He had a limited number of STR's (9 or so).  I don't know if its right for long duration estimates but it didn't work on the Ian Cam of Clan Gregor where i was sure I was working with unique mutational events.

One of the big problems I see is the very slow mutators, they're better for longer time estimates, but it's hard to get enough data.

The other option to simple data collecting is to look at what Klyosov and his chemical kinetics is saying.  I've observed that he is a bit "crusty", but I believe even Nordtvedt agreed with some of his work?

I'm not sure we can even trust the ISOOG tree.  If some of the early lines had a bottleneck, and it appears they did, and if we don't have some of the early SNP's from A or B, then we don't have a real good leg to stand on to anchor the SNP line (s).
Logged
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #11 on: June 10, 2012, 02:35:57 PM »

There is a diversity cline in R-M269 as a whole from SE to NW:

R-M269 Diversity

That holds true whether one uses Zhivotovsky or germ line.

Some argue that R-M269 diversity as a whole does not matter, that one must separate the western R-L51 clades from the eastern xL51 clades and look at them as almost two unrelated things; but that seems like special pleading designed to salvage a Paleolithic or Mesolithic arrival for R1b in Europe.

A more parsimonious interpretation of R-M269 diversity, it seems to me, is that R-M269 moved into and across Europe from SE to NW in wave fashion. L51 and subsequently L11 and its offshoots surfed the outer edge of the expanding wave. L11's two major subdivisions, P312 and U106, were probably born somewhere in Central Europe and expanded from there.
Logged

Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #12 on: June 10, 2012, 03:04:15 PM »

There is a discussion on rootsweb re: Does the modal value change with time.  In fact, Chandler steps up and says it is a random walk process.  

I question that assertion.  Please go to: www.freepages.genealogy.rootsweb.ancestry.com/~geneticgenealogy/yfreq.html. and replot the modals for the 7 Hgs as a function of dys loci.

What you will observe is that the modal is fairly constant for most Hgs, with some multisteps affecting the process.  This suggests to me that a random walk model is not appropriate?

Are you replying on Rootsweb?
No.  I mailed the person asking the questions the data set.

I recommend you put your questions out directly to the people you disagree with. To simply post "in absentia" comments here is of no value.  Why should we think your points carry any value if you won't argue your position directly and publicly?

I don't really care, other than I don't want a newbie to read your comments without understanding the circumstances.
 This is a lot of bull mike.  You have your agenda, which I don't happen to agree with and would also be concerned if a newbie read it.  This will sort itself out in time.  Going against the prevailing wind is slow and tedious and admittedly frustrating.  This is your board.  Do with it as you like.

What do you mean "this is your board" ?   I have nothing to do with this WWF forum other than as a poster...  I'm just a lowly poster.

BTW, since you can read minds, what is my "agenda?"  ]
« Last Edit: June 10, 2012, 03:13:05 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #13 on: June 10, 2012, 03:09:41 PM »

I really haven't gotten into trying to predict how people travelled yet.  It depends which time frame we are talking about.  Aurignacian, 40K BP?  Or does everything begin with the Holocene, which is better understood?  Whoever lived in Europe before the last Ice Age was displaced and I suspect he went Eastward and Southwards.  Some followed the game across the tundra to Asia, others went back to the Refugia.

Any evidence of pre LGM in northern Europe is gone.  The Holocene is basically all we have and their are few coastal sites left because the ocean has risen at least 60 Meters since the LGM.  Not only the coastline but the climate and aridness has changed.  10K years ago North Africa started changing into a fertile area with water and game available.  This must have caused some displacements?  Southern Europe wasn't as hospitable as it is today, wheras about 9-11K Bp northern Europe was much warmer than it is now.

How one, using STR's, creates a readable story out of that scenario is pretty questionable.

One last issue is the great flood.  Where was it, what did it effect.  Just Noah and his Ark? It seems to have been very severe if you believe the Lake Agassiz story?

What it boils down to is we really only have a benign situation for the last 4 to 6K years upon which to base our analysis.  Anything prior to that is one hell of a guess.  JMHO.

Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #14 on: June 10, 2012, 03:11:23 PM »

I must confess I am losing my confidence in age estimates based on STRs. I think they have some value, but it's limited. The problem is that there is nothing in y-dna itself that really tells you how old it is.....

I think it was Ken who described the use of multiple rough Y STR molecular clocks in a statistically aggregate way is really akin to what Carbon-14 dating is too.

It's just that your Y STR composite clocks are at very early stage of development with lots of work to be done yet on mutation rates, etc.

Both ancient DNA and and using modern DNA have a common problem...   we are long way from having representative data.
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
rms2
Board Moderator
Guru
*****
Offline Offline

Posts: 5023


« Reply #15 on: June 10, 2012, 03:52:21 PM »

As I've said before, I don't believe R-M269 was in Europe (well, western Europe at least) before the Neolithic. The main argument that it was stems from the modern frequency of R-M269, i.e., the reason there is so much R-M269 in Europe today is because it got there very early.  But the STR diversity in western Europe is low. Genetic bottlenecks are supposed to account for that, and maybe they do, but they devastate the argument from modern frequency. I've said this elsewhere before, but it bears repeating. If R-M269 could recover from a bottleneck or series of bottlenecks so severe that it or they could create the "illusion" that R-M269 In Europe is much younger than it actually is, then the argument from frequency is gone. Whether R-M269 recovered after bottlenecks or arrived in Europe late, it's all the same: it achieved its success since that time and not as a consequence of steady, continuous growth since the Paleolithic or Mesolithic. So, an early arrival is either wishful thinking or a non sequitur. It could be true, but high modern frequency is not evidence of it, given low haplotype diversity.

Thus far, the aDNA evidence has not supported an early arrival for R-M269 either. Of course, that could all change tomorrow with one Mesolithic or Paleolithic European R-M269 result. I'm betting that won't happen, however.
« Last Edit: June 10, 2012, 03:56:33 PM by rms2 » Logged

Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #16 on: June 10, 2012, 08:41:45 PM »

...Finally, I would recommend everyone read N. Talebs book, the Black Swan.  It descredits the use of the Gaussian in the Financial industry and points out many errors in logic based on the premise that "everything can be described by a Gaussian curve"  JMHO>

There is plenty to read out there.  Perhaps, rather than recommending books to read or making general references to allele distribution frequencies table, you could explicitly describe your point.

« Last Edit: June 10, 2012, 08:42:05 PM by Mikewww » Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #17 on: June 11, 2012, 07:35:57 AM »

Talebs book is all about the "land of extemistan" where many physical processes exist.  Its the region beyond the 3 sigma point of the Gaussian.  He gives an example of a turkey living for some 200+ days, expecting to live the day before thanksgiving, and then is killed for the harvest table.  Obviously, the turkey had no idea why he had such a benign life.  For the turkey, this was a Black Swan.  I have showed on a separate thread that the occurrence of a mutation exceeds the three sigma value of the Gaussian(in a plot of # of mutations vs births), further all the different dys loci occurrences occur further out on the curve.  Ergo, the mutational process is not described by the gaussian. (This is amplified by the fact that all mutational step sizes are not equal).

The Y STR frequency plot presents many interesting points.  Dys loci mutations do not follow a random walk, they are constrained.  The faster mutators, such as CDYa,b approach a uniform distribution across 3 to 4 allele values.  Apparent multistep mutations can significantly change the modal value and in the case of 388, increase the apparent number of mutations for say I, J hgs.  Its also seems that the modal itself doesn't follow a random walk pattern, but is more likely to stay the same across the  7 hgs presented (E3a,E3b,G, I , J, R1a, R1b).  This data set puts in question many of the assertions about the property of STR's.

My major point is that we should be creating information similar to this STR frequency plot as we accumulate  more data and try to intrepret  the results, not creating theories.  Klyosov has asserted that the use of 22 slower mutators, having an average mutational period of some 4250 years should be used to estimate long time frames, greater than 10K years.  Can we create a new estimate of Hg age using this assumption.  For instance, I showed (having 8 mutations) that my haplotype is separated from M269/P25 by 40 K years.  Does that make any sense?  Does the data that Machiavelli has collected provide any insight to the age of R1b?

I know that certain folks have their minds made up about how the mutational process evolves.  I don't.  But, for right now, I don't ascribe correctness to any of the theories we have.  Will we ever have proof?  I think we are going to have to look underwater.  JMHO.
« Last Edit: June 11, 2012, 07:37:31 AM by ironroad41 » Logged
Jean M
Guru
*****
Offline Offline

Posts: 1253


« Reply #18 on: June 11, 2012, 07:40:28 AM »

 Does the data that Machiavelli has collected provide any insight to the age of R1b?

Was that a Freudian slip? :)
Logged
Jean M
Guru
*****
Offline Offline

Posts: 1253


« Reply #19 on: June 11, 2012, 07:53:35 AM »

@ ironroad41

You seem to be asking "Is there any way to make estimated ages older?" Even I can answer that one. Yes of course. You just choose a method which will produce estimates of the desired age.

You may then have trouble getting anyone else interested in your results. That's another story entirely from the mechanics of number-crunching. Estimates of mutation rates have to be calibrated by a method independent of the actual business of creating them. The method could be genealogical or by radio-carbon dating of bones which are DNA tested, or anything else which gives you an external clock.  Without that calibration, age estimates are worthless.


 
« Last Edit: June 11, 2012, 07:54:32 AM by Jean M » Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #20 on: June 11, 2012, 08:08:04 AM »

Getting older age estimates isn't my objective.  Getting more accurate age estimates is.  The existing set of knowledge has been constructed using one theory of the mutational process.  Like any institution/process it then has a life of its own and desires self-preservation.

Dienekes has flat said that STR's don't work, but I don't agree with that - yet.  You are right, Klyosovs work would probably stretch the estimates and I know it won't agree with the common accepted knowledge, so what does that really mean?  Who is right?

Either the mutational process is self-consistent in some sense or it isn't.  I think that everyone is looking for some other technology to step up and help us calibrate, beyond the meosis data we are accumulating.  I think one of the major issues we are faced with is that when all the current data is put into its little piles by SNP, we are still faced with some pretty tough questions.  How does a bottleneck affect the data?  Are haplotypes with several slow mutations old or simply the result of a random process?  Does the fact that population growth has taken off over the last 4K to 6K years affect our TMRCA estimates. etc., etc.
Logged
Jean M
Guru
*****
Offline Offline

Posts: 1253


« Reply #21 on: June 11, 2012, 08:11:11 AM »

Either the mutational process is self-consistent in some sense or it isn't.  I think that everyone is looking for some other technology to step up and help us calibrate,

It is not a question of "helping" to calibrate. Calibration must be external. That is the only way to calibrate. The biology does not come with its own absolute clock. The only way to test the accuracy of results is against some external evidence.

I can recommend a good textbook: M.A. Jobling, M.E. Hurles and C. Tyler-Smith, Human Evolutionary Genetics: Origins, Peoples and Disease (2004). 
« Last Edit: June 11, 2012, 08:21:56 AM by Jean M » Logged
Mike Walsh
Guru
*****
Offline Offline

Posts: 2964


WWW
« Reply #22 on: June 11, 2012, 08:17:04 AM »

.... Can we create a new estimate of Hg age using this assumption.  For instance, I showed (having 8 mutations) that my haplotype is separated from M269/P25 by 40 K years.  Does that make any sense? ....

You use the word "show" and I equate that to "demonstrate."  These are fairly assertive term indicating you have provided some level of proof of what you are saying.  I missed that evidence and logic or I guess I maybe just misunderstood it or forgot it.

How have you shown that your haplotype is separated from M269 or P25 by 40K years and which is it? M269 or P25?
Logged

R1b-L21>L513(DF1)>S6365>L705.2(&CTS11744,CTS6621)
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #23 on: June 11, 2012, 08:48:11 AM »

Either the mutational process is self-consistent in some sense or it isn't.  I think that everyone is looking for some other technology to step up and help us calibrate,

It is not a question of "helping" to calibrate. Calibration must be external. That is the only way to calibrate. The biology does not come with its own absolute clock. The only way to test the accuracy of results is against some external evidence.

I can recommend a good textbook: M.A. Jobling, M.E. Hurles and C. Tyler-Smith, Human Evolutionary Genetics: Origins, Peoples and Disease (2004). 
  I have that book as a reference.  for what we are discussing here, it is out of date. JMHO.
Logged
ironroad41
Old Hand
****
Offline Offline

Posts: 219


« Reply #24 on: June 11, 2012, 08:52:50 AM »

.... Can we create a new estimate of Hg age using this assumption.  For instance, I showed (having 8 mutations) that my haplotype is separated from M269/P25 by 40 K years.  Does that make any sense? ....

You use the word "show" and I equate that to "demonstrate."  These are fairly assertive term indicating you have provided some level of proof of what you are saying.  I missed that evidence and logic or I guess I maybe just misunderstood it or forgot it.

How have you shown that your haplotype is separated from M269 or P25 by 40K years and which is it? M269 or P25?
  If you accept Klyosovs assertions of his approach, then I have demonstrated that fact.

I used the set of 22 dys loci values klyosov calls R1b1.  This is in his Out of AFrica paper published on-line May 2012.
Logged
Pages: [1] 2 3 Go Up Print 
« previous next »
Jump to:  


SEO light theme by © Mustang forums. Powered by SMF 1.1.13 | SMF © 2006-2011, Simple Machines LLC

Page created in 0.11 seconds with 19 queries.